The signalling of internal states of each cell potentially facilitates co-ordinated doing — if cells are inclined to react to each other’s expressed proteins in conditioned circumstances.
All the examples are eukaryotes, rather than prokaryotes (bacteria), so this system probably emerged with the evolution of the eukaryotic cell.
A well known example that fits this scenario is the slime mould, whose life cycle alternates between unicellularity and multicellularity. As Dawkins (2004: 416-7) explains:
Cellular slime moulds are social amoebas. They literally blur the distinction between a social group of individuals and a single multicellular individual…[Separate amoebas] converge on aggregation centres, from which chemical attractants radiate outwards. As more and more amoebas stream in on an attraction centre, the more attractive it becomes, because more of the chemical is released…Eventually the amoebas in each major attraction centre unite their bodies to form a single multicellular mass, which then elongates into a multicellular ‘slug.
Slightly more integrated is a sponge-like colonial organism, Proterospongia, which is a population of individual single-celled organisms called choanoflagellates (Dawkins 2004: 400-9).
And more integrated still are the sponges (porifera), which you see pictured on the slide, although these may be the borderline case.
As Dawkins (2004: 400-1) points out:
Sponges don’t have embryology [like other multi-celled animals]. Instead they self-assemble — each of their totipotent cells has an affinity for hooking up to other cells, as though they were autonomous protozoa with sociable tendencies…
Now, it is the expression of genes as proteins that provides the basis for the emergence of intercellular semiotic systems, and through them, (“true”) multicellularity, as we will now see.